Defence and National Security Calibration
Public overview of UAIF-CAL-DEFENCE-v1.0; the operational package is controlled.
Purpose and scope
Post-incident classification and routing only; weapon legality, targeting, lethality and intelligence operations are excluded.
The package calibrates UAIF classification for sector context without replacing the UAIF Technical Specification or applicable law.
Status and validation boundary
Version 1.0 is provisional pending empirical inter-rater validation. The source document states that provisional weights must not be used as the sole basis for regulatory reporting, certification or consequential decisions.
Public information
- Document reference: UAIF-CAL-DEFENCE-v1.0
- Version: v1.0 provisional
- Public scope and exclusions
- Method based on public regulatory material, research and recognised severity frameworks
- Known validation and legal-review limitations
Sector risk context
Defence and national-security AI incidents carry disproportionate strategic and human consequences. AI is deployed across intelligence/surveillance/reconnaissance (ISR), command-and-control support, cyber defence, logistics, geospatial intelligence, and early-warning systems. This package classifies and routes incidents in such systems — failures, malfunctions, compromises, or misclassifications — after they occur. Dominant harm vectors are physical harm (including civilian harm), security/integrity (mission and information-security compromise expressing strategic and rights impacts), and privacy. The package does not model, recommend, or optimise any offensive capability.
1.1 Applicable Regulations
• Geneva Conventions; Additional Protocols I & II; international humanitarian law (IHL) / law of armed conflict (LOAC) [T1]
• EU AI Act [T1]
• Export controls: ITAR; EAR; EU Dual-Use Regulation; India SCOMET [T1]
• NATO Responsible AI Principles; UN GGE on LAWS recommendations; NIST AI RMF; ISO/IEC 23894 [T2]
Extracted from the source sector profile. Detailed calibration and operational material remains controlled.
Public sector orientation
Representative classification concerns include human oversight, security integrity, escalation, protected information, mission impact and applicable humanitarian constraints. These themes help teams scope incident intake and analysis; they are not calibration weights, legal conclusions or an exhaustive risk list.
Use with the UAIF layers
- Establish incident identity, provenance and remediation state.
- Separate cause, manifestation, acute harm and chronic-harm proxies.
- Apply severity and context logic with the provisional calibration status visible.
- Record sector and jurisdiction metadata for human review and routing.
- Preserve uncertainty, contrary evidence and the version of every referenced artifact.
Controlled resource
The detailed package, calibration weights, scoring implementation, evidence logic and maintenance material are confidential commercial resources governed by UAIF-LIC-003 and applicable GEL terms.
Notably Absent
No empirical validation result, regulatory approval, certification status, universal sector applicability or legal-reporting determination is claimed.